Adaptive Self-Tuning Wavelet Neural Network Controller for a Proton Exchange Membrane Fuel Cell

Abstract

Movable power sources of Proton Exchange Membrane Fuel Cells (PEMFC) are the key study direction in the existing Fuel Fells (FC) research field. In order to improve the performance of the PEMFC, extend its life, increase its safety and lower the system costs, it should be controlled effectively. Among various parameters to control, the electrolyte membrane moisture, affecting the performance of the PEMFC, is an important controlled variable. PEMFC presents a number of remarkable control demands. Consequently, traditional control methods cannot be applied to control PEMFC due to the imprecision, uncertainty and partial truth, and its intrinsic nonlinear characteristics. Hence, self-tuning adaptive control based on neural networks is an attractive method to control a PEMFC. In this chapter, two major wavelet neural network-based control structures will be presented. These controllers are based on the combination of a single layer feed-forward neural network with hidden nodes of adaptive wavelet functions, and an Infinite Impulse Response (IIR) recurrent network. The IIR is cascaded to the neural network to provide a new network leading to improvements in the speed of learning. These particular neural controllers assume a certain model structure to identify the system dynamics of the unknown plant approximately and generate the control signal. The first one is a self-tuning wavelet neural network controller and the second one is an adaptive self-tuning Proportional-Integral- Derivative (PID) controller using a wavelet neural network. The advantages and disadvantages of the two proposed controllers will be discussed and illustrated in this chapter. The well-known Multi Layer Perceptron (MLP) with traditional Back-Propagation training (BPP) will also be included in the control design for a base line comparison. The proposed controllers are studied in three situations: without noise, with measurement input noise, and with disturbance output noise. Finally, the results of the performance of the new controllers are compared with a multilayer perceptron network, proving a more precise modeling and control of PEMFC.